Subtropical boundary layer cloud feedbacks on climate change – the CGILS LES/SCM column modeling intercomparison

Subtropical boundary-layer cloud feedbacks are a profound source of disagreement between global climate models, and are the single largest uncertainty in predicting the sensitivity of global-mean surface air temperature to greenhouse gas increases. The intermodel range in these feedbacks has not greatly decreased over the past two decades, even as climate models have gradually improved their simulations of clouds in the current climate. One issue has been a lack of understanding of what are the key physical mechanisms for subtropical boundary-layer cloud response to a climate change, both in reality and in climate models. Mechanisms can be classified as ‘dynamic', involving rearrangement of the climatological general circulation (e. g. poleward migration of the storm tracks) and ‘thermodynamic' (e. g. changed free-tropospheric static stability or relative humidity). Climate models disagree particular about the effects of thermodynamic changes in a future climate.

This suggests that much can be learned by a column modeling approach in which different column and LES models of a cloud-topped boundary layer are all run with the same control and perturbed surface and free-tropospheric conditions, and the cloud responses are compared. The CFMIP/GCSS Intercomparison of LES and SCMs (CGILS) is an international intercomparison study started in 2009 to do this. The focus has been on three locations (characterized by stratocumulus, transitional, and cumulus-topped boundary layers) in summertime on a NE Pacific ‘GPCI' cross-section running southwestward from California past Hawaii. Boundary conditions and advective tendencies were extracted from reanalysis and climate change perturbations corresponding to a uniform 2 K SST increase (but no CO2 change) were extracted as averages across a representative set of climate model simulations.

Single-column versions of most major climate models as well as five LES have been run with these forcings. The diversity of responses is even larger than for the full climate models. The LES and most SCMs show a consistent deepening of the boundary layer in the warmer climate due to reduced subsidence. In the LES, this results in a slight cloud increase, but in many SCMs it results in a cloud decrease because the deeper boundary layer supports more parameterized shallow cumulus convection and less stratocumulus. However, higher vertical resolution is needed for the LES to be trustworthy. The CGILS approach shows promise for better understanding and improving cloud feedbacks in single-column versions of climate models, and hopefully thereby decreasing the range of simulated cloud feedbacks in their global analogues.